The Penn State Worry Questionnaire: psychometric properties of the Korean version

Abstract

The factor structure and concurrent validity of the Penn State Worry Questionnaire (PSWQ) were examined in two college student samples in Korea. We demonstrated method effects due to the inclusion of negatively keyed items. Confirmatory factor analyses supported the single-factor model with method factor. This indicates that the Korean version of the PSWQ (K-PSWQ) can be contaminated by method effects and response patterns are different between positively worded and reverse-scored items. Thus, the relevance of negatively worded items in the K-PSWQ is questioned.

Penn State Worry Questionnaire: structure and psychometric properties of the Chinese version

“…This is consistent with some previous studies, which found that the one-factor-withmethod-effect model provided a better fit than the two-factor structure (Brown, 2003;Hazlett-Stevens et al, 2004;Lim et al, 2008). In the method-effect model, the covariance of the 5 reversed items is assumed to be due to substantively irrelevant method effects, not some latent factors.…”

“…Later, Hazlett-Stevens et al(2004) found that the five reverse-direction PSWQ items do not make up a separate worry construct, but rather, contain an underlying factor of negative wording method. This result has recently been replicated in a sample of Korean college students (Lim et al, 2008). However, a flaw in the aforementioned studies (Brown, 2003;Hazlett-Stevens et al, 2004;Lim et al, 2008) is that the two-factor model and one-factor model with method effect have indistinguishable goodness of fit.…”

“…This result has recently been replicated in a sample of Korean college students (Lim et al, 2008). However, a flaw in the aforementioned studies (Brown, 2003;Hazlett-Stevens et al, 2004;Lim et al, 2008) is that the two-factor model and one-factor model with method effect have indistinguishable goodness of fit. The decision to accept the one-factor model with method effect over the two-factor model was made without due consideration of the statistical result of model comparison.…”

“…In the method-effect model, the covariance of the 5 reversed items is assumed to be due to substantively irrelevant method effects, not some latent factors. Although this assumption of method effect has been proven in prior studies (Brown, 2003;Hazlett-Stevens et al, 2004;Lim et al, 2008), the strategy of using model comparison as a way to evaluate the best acceptable model is rather novel. Our finding supports that, in a Chinese non-clinical sample, it is not necessary to split the Ch-PSWQ into two distinct factors named "engagement of worry" and "absence of worry."…”

“…Model 2 was the two-factor solution, with the 11 positively worded items and the 5 negatively worded items representing the two indicators (Beck et al, 1995;Fresco et al, 2002;van Rijsoort et al, 1999;Stöber, 1995). Model 3 was the one factor with method effect solution suggested by Brown (2003), and subsequently repeated by Hazlett-Stevens et al(2004) and Lim et al(2008). As the goodness-of-fit indices shown in Table 2, Models 2 and 3 are superior to Model 1, and while both Models 2 and 3 are acceptable, Model 3 is slightly better.…”